Tin in mantle-derived rocks: Constraints on Earth evolution

Tin in mantle-derived rocks: Constraints on Earth evolution

OOlb-7037/93/~6.~ + .oo Tin in mantle-derived rocks: Constraints on Earth evolution K. P. JOCHUM, A. W. HOFMA~N, and H. M. SEUFERT Max-Planck-Instit...

1MB Sizes 55 Downloads 67 Views

OOlb-7037/93/~6.~

+ .oo

Tin in mantle-derived rocks: Constraints on Earth evolution K. P. JOCHUM, A. W. HOFMA~N, and H. M. SEUFERT Max-Planck-Institut (Received

February

Nr Chemie, Saarstrasse 23, D-55122 Mainz, Germany

I 1, 1992: accepted in revised form March 1, 1993)

Abstract-We have analyzed tin together with other trace elements in a variety of mantle-derived rocks by isotope dilution-spark source mass spectrometry. Tin concentrations in oceanic basalts (mid-ocean ridge basalts, or MORB; oceanic island basalts, or OIB; island arc volcanics, or IAV) vary from 0.4 to 6 ppm. Tin is a moderately siderophile element which behaves like the moderately incompatible and 1~thophile element Sm during igneous processes in the mantle. The Sn/Sm ratio is therefore similar in various reservoirs of the silicate portion of the Earth at the present day. To investigate possible secular variations of siderophile/chalcophile element abundances, we have also analyzed komatiites and basalts with ages ranging from Archean to Tertiary. Rocks having ages up to 3.4 Ga yield Sn/Sm ratios comparable to those of the oceanic basalts. These results indicate that core growth during the last 3.4 Ga was negligibly small. In addition, the Sn/Sm ratios in different mantle reservoirs are identical within the uncertainty of the data scatter. This indicates that tin was homogenized within the mantle after core formation. We conclude that the value of Sn/Sm = 0.32 found in mantlederived rocks is representative of the primitive mantle because the estimated continental-c~stal abundances of Sn and Sm yield a similar SnfSm ratio. This results in Sn = 0.12 + 0.01 ppm for the primitive mantle abundance, in agreement with that determined from fertile peridotite xenoliths (0.15 + 0.02 ppm). To determine an absolute depletion factor for the volatile siderophile element tin in the primitive mantle, we have also analyzed four carbonaceous chondrites, including the CI chondrites Orgueil and Ivuna. We obtain a CI-value of tin of I .62 Z?Z 0.03 ppm. Tin is depleted in the Earth’s primitive mantle bv a factor of 33 $: 3: the volatility corrected depletion factor is similar to other moderately siderophile elements ( NEWSOM, 1990). _ dilution using the spiked graphite method (ID-SSMS). and c~ib~tion with relative sensitivity factors (RSF). To measure tin by isotope dilution, about 50 mg ofsample powder were mixed with a spiked graphite containing ‘19Snspike in an agate mill. The mixing ratio varied from I: I to 2: 1,depending on the tin content. Concentrations were determined using both “‘Sn/“‘Sn and “8Sn/“9Sn ratios in the spiked samples. In most cases, “‘SK “*Sn, and ‘19Sn isotopes were interference-free; possible interferences of molecules could generally be avoided by the high mass resolution (about 5000) of the spectrometer. Replicate analyses showed that the precision of the ID-SSMS measurements is better than 7%, if spiking was optimal (isotope ratios close to unity). Tin concentrations were also determined conventionally by measuring Sn/Nd ratios, calibrated with a RSF for tin (which varies from 2.03 to 1.4 I for sample/graphite mixing ratios from 1:I to 2: I ) and usine. Nd as an internal standard element (determined bv ID-SSMS). The-precision of this latter method is lower (about l&15%) than the more direct isotope dilution measurements. In most cases, both independent methods for tin (ID and RSF) agree within the error limits, demonstrating the reliability of the analyses. The detection limit for tin is about 0.01 ppm, allowing its determination in a wide variety of rock types. The accuracy of the tin analyses can be tested by comparing the results of international standard rocks with literature values. Table 1 and Fig. I show our data together with the working values by GOVINDARAJU( 1989). The working values are of varying quality: only three of them ( W- 1,SDC- 1, RGM- 1) are recommended (probably close to true) values, most are proposed values, and three are magnitudes (information values). The agreement is better than 25% for most of the samples with no systematic differences apparent. The only exceptions to this are ( 1) where the working values are only magnitudes, and (2) the data for the peridotite standard PCC- 1.

TIN BELONGSTO A GROUP of elements which are called moderately siderophile. Some of the elements of this group, which also includes W. MO, and Pb, have particularly interesting geochemical properties because they have participated in both of the major differentiation processes of the Earth, the segregation of the core, and the segregation of the crust. During these processes, their “doubly incompatible” geochemical properties cause them to be partitioned into metal/sulfide as well as into silicate liquids. Thus, the siderophile/chalcophile character accounts for their partitioning into the core and corresponding bulk depletion in the silicate portion of the Earth. Subsequently, these elements are also enriched in the crust, which is formed by the segregation of silicate liquids. The behaviour of W, MO, and Pb during basalt genesis has already been investigated by NEWSOM and PALME (19841, NEWSOM et al. f 1986), and SIMS et al. (1990). Until now, there have been few data for tin abundances in mantle-derived rocks, mainly because of analytical difficulties. In this paper, we have used spark source mass spectrometry (SSMS) to determine tin abundances in modern basalts (MORB, OIB, and IAV) and in igneous rocks of different ages. This allows us to place important constraints on core formation, accretion, and evolution models for the Earth. ANALYTICAL

TECHNIQUE

Spark source mass spectroscopy has been used for the determination of tin (in addition to about thirty trace elements). The SSMS technique has been described in detail by JOCHUMet al. (1988). For quantitative analysis. two different methods were applied: isotope

SAMPLES

A wide variety of samples of differing provenance and tectonic setting were investigated in this study. The rock types include basalts, 3585

3586

K. P. Jochum, A. W. Hofmann, and H. M. Seufert

Table 1. Sn concentrations (ppm) in international standard rocks determined by SSMS. The data are compared with working values by GOVINDARAJU (1969) (data u&&& are recommended values; other values are proposed except those enclosed in brackets which are magnitudes).

Geostandard

SSMS

BCR-1

2.33 2.95 4.61 3.94 0.54 2.39 1.76 4.9 5.06 0.23 2.20 3.59 2.39 1.0 4.69 3.6 2.0 0.43

W-l

AGV-1 w-2 BIA-1 BHVO-1 BE-N NIM-G STM-1 NIM-N MRG-1 SDC-1 QLO-1 PCC-1 RGM-1 G-l G-2 JGb-1

Working values 2.7 zz 4.2 (0.7) 2.1

La

Hf

Sm Sn Eu Gd

Y

Ho Er Yb Lu

FIG. 2. Primitive-mantle-normalized concentrations of tin and some immobile trace elements from samples of core (“fresh appearing interior”) and margin (“presumably weathered outer zone”) of the oceanic basalt DH 08 (JOCHUM and VERMA, 1993).

4 6.6 (1) (3.6) Q 2.3 1.6

published by HOFMANNet al. ( 1986). ITO et al. ( 1987), JOCHUMet al. (1983), and NEWSOMet al. (1986).

&.l 3.2 (1.6) 0.36

Oceanic Island Basal&

komatiites, and peridotites. Additionally, several carbonaceous chondrites were analyzed for comparison. Brief descriptions of the analyzed samples are given in the following discussion. Mid-Ocean

Pr Nd Zr

Ridge Basalts

Dredged mid-ocean ridge basalts (MORB) samples from different locations of the Atlantic, the Pacific, and the Indian oceans were analyzed. Most of these were obtained from the basalt glass collection of the Smithsonian Institution. Investigations on altered MORB samples (JOCHUMand VERMA, 1993) have shown that seawater alteration may enrich or deplete tin concentrations by up to a factor of three (Fig. 2). To avoid seawater contamination, only clean glass fragments from the samples, carefully hand-picked and with no traces of palagonite, were selected. Descriptions of the hand-picking technique have been given in detail by IT0 et al. (1987). Most samples belong to the depleted type (N-type) MORB. Data on location, tectonic setting, depth, and major and trace-element compositions have been reported by MELSONet al. ( 1977) and PUCHELTand EMMERMANN( 1983); locations are also listed in Table 2a. Isotope ratios and trace-element data have been

The samples come from the Hawaiian Islands, Samoa, St. Helena, Tubuai, Azores, Society Islands, Tristan da Cunha, Cough Island, Comores, Reunion, and Galapagos. The Hawaiian Islands were studied most extensively. Tin was determined in rock samples ofdifferent types from nearly all volcanoes of the Hawaiian chain (Table 2~). Most oceanic island basalts (OIB) samples (mainly tholeiites and alkali basalts) were fresh and taken from historic eruptions where possible. Locations, sample descriptions, major element, trace-element, and isotopic data are given by HEDGEet al. ( 1972), HEGNER et al. (1986), HOFMANNet al. (1986), JOCHUMet al. (1983), NEWSOMet al. (1986), STILLEet al. (1983. 1986), WEST et al. (1987), WHITE et al. (1979), and WRIGHTet al. (1975). Island Arc Volcanics The samples come from eight islands of the Lesser Antilles island arc. Brief sample descriptions, isotope ratios, and trace-element data are published by WHITE and DUPRE ( I986 ). Komatiites and Basal& (With Ages Ranging from 3.4 Ga to Tertiary)

Archean to Tertiary komatiites and basalts were analyzed. The Archean rocks fall into two age groups: 3.4 Ga old rocks from Barberton (S. Africa) and Pilbara (Australia); and 2.7 Ga old rocks from Kambalda (Australia), Abitibi (Canada), and Belingwe (Zimbabwe). Proterozoic samples were taken from the Ottawa Islands (Canada) and are about 1.9 Ga old. The youngest samples of this type are Tertiary komatiites and basalts from the Gorgona Island (Colombia). Other trace-element data and the sources for petrology and major element data can be found in JOCHUM et al. ( 199 I ) Spine1 Peridotite Xenoliths Tin and other trace-element data were published previously by et al. ( 1989) for five spine1 peridotite xenoliths. Locations for these samples are San Carlos (Arizona, USA), Kilbourne Hole and Potrillo (New Mexico, USA), Landoz ( MassifCentral, France), and Dreiser Weiher (Eifel, Germany). We have analyzed four additional peridotite xenoliths from southeastern Australia as part of this present study. The sample descriptions, isotopic compositions, and trace-element data are given in MCDONOUGH and MCCULLOCH ( 1987). JOCHUM

-.J Ll

2

4 6 8 Working value ( ppm )

10

FIG. 1. Comparison of SSMS analyses of tin in standard rocks with working values by GOVINDARAJU (1989). Our data agree within 25% (shaded area) with the reference values, whereas the data of TAYLOR and MCLENNAN( 1983) are systematically higher.

Carbonaceous

Chondrites

In order to determine accurately both the solar-system abundance oftin and the depletion factor of tin in the primitive mantle, we have

Tin in mantle-derived rocks Table 2a. Analyttcal results of MORB Sample ATLANTlCOCEAN VG 962 VG 367 VG 965 VG 200 VG 744 VG 205 VG 937 VG 941 P6909-28B

GS7309-75 VG 198 VG 192 PAC/F/COCEAN VG 768 010-l 010-2 DlO-3 R3-3-030 R3-3.DlO VG 973 KlOA-D33A VG 798 K62A-Dl43G VG 1770 K73A-D123H /~DlAN OCEAN VG 1583 VG 3095

Location

3587

(concentrations in ppm) Sn

Nd

Sm

Eu

Zr

Hf

SniSm

4.30 7.26 8.29 17.9 12.2 13.8 10.8 14.4 12.3 11.9 13.5 13.5

1.57 2.56 2.74 3.82 3.88 4.24 3.70 4.39 4 04 3.68 4.20 4.54

0 611 1.09 1.02 137 1.35 1.48 1.26 1.66 1.41 132 1.49 152

39 4 84.3 81.4 703 108 139 109 134 85.4 78.0 124 135

105 2.26 2.33 2.80 2.96 3.36 3.23 3 85 2.66 2 50 3.38 3.86

0.501 0.395 0.547 0 390 0.314 0 406 0 381 0 296 0.309 0.280 0.445 0.436

46.4"N 130.2OW 44.7"N 130.3"W 447ON 130.3*W 44.7"N 130.3"W 13.8"N 104.l"W 12.l"N 103.8"W 1.5"N 101.4°W 20.4"s 114.O"W 31.0"s 113.lOW 2.6"N 95.3-W l.O"N a5.vw 1.5"N 85.1"W

5.89 11.3 16.0 12.3 13.7 1.29 141 1 .33 1 27 8.35 2.64 22.8 1.19 9.72 1.31 10.7 1.27 14.1 0.627 4.81

2.16 4.02 5.60 4.31 4.54 4.38 2.73 7.89 3.08 3.59 3.27 2.01

0.897 1.52 1.98 1.35 1.58 1.54 1.06 2.50 1.15 124 1.41 0.822

57.5 104 134 123 103 113 75 3 227 120 105 118 42.0

1.73 3.24 4.04 3.85 3.03 2 90 212 7.15 3.24 2.90 3.10 1.40

0.486 0.254 0.257 0.267 0 284 0.304 0.465 0.335 0.386 0.365 0.388 0.312

5.4"N 68.7% 25.O"S 70.O"E

1.65

4.51 3 29

1.55 1.22

132 96.7

3.96

1.10

0.366 0.334

70.2"N 15.3OW 52.7"N 34.9"W 49 8"N 28.7"W 43.O"N 29.2"W 25.4"N 45.3"W 22.9"N 13.5-W 22.2"N 45.O"W 22.2"N 45.3"W 6.O"N 33.3"W 0.6"s 16.1°W 21.9"s 11.9ow 21.9% 11.8"W

0.786 1.01 1.50 1.49 1.22 1.72 1.41 1.30 1.25 1.03 1.07 1.98 1.05 1.02 1.44 1.15

12.9

9.10

Table 2b. Analytical resultsof OIB (concentrations in ppm) Sample

Sll

Azcx?E.s 2.51 F-33 3.41 SM-12 SM-6 3.52 1.84 P-21 GALAPAGOS 1.86 Sant. E-20 1.94 Isab.ZOE36 S7YfELENA 2.05 2862 3.11 2926 2.51 2928 4.25 102 1.26 SC66 2.63 SC56 1.14 SC38 2.20 SC74 TUB/JAI 2.43 5433 5.08 TU9 6.10 1052 SAMOANISLANDS VP-1 2.40 VP-2 2.30 2.87 UPO-7 82-MT-15 2.68 JKU-1 3.36 SAV-1 211 TRISTAN DA CUNHA TR-1 3.27 TR-4 3.13 TR-5 3.63 TR-6 3.70 GOUGHISLAND G-8 3 09 SOClETY /SLA NDS Hauh 769 5.53 Tahi 343 3.48 Tahi 351 2.69 Tahaa 73.185 6.06 AJOAJ21-9 f?EUN/C#V RE24-1

Sm

ELI

Zr

Hf

35.6 54.0 50.2 27.1

7.07 104 10.5 5 82

2.28 3.21 3.11 2.07

213 296 318 166

6.06 8.17 8.75 4.88

0.355 0.328 0.335 0 316

20.4 22.3

4.65 5 17

1.59 1.85

167 140

4.12 3.69

0.400 0.375

31.2 49.5 35.6 67.8 25.8 49.1 22 0 43.7

5 47 10.2 7 60 121 5.36 10.9 4.58 7.89

1.78 3.28 2.41 3.60 1.67 2.93 1.34 2.52

170 298 256

4.33 7.56 6.42

189 354 138 290

4.54 7.73 3.41 6.25

0.375 0.305 0.330 0.351 0.235 Ct.241 0 249 0.279

57.6 127 102

11.1 18.5 14.3

3.29 4.76 4.00

301 1050 1040

33.6 27.8 75.9 38.2 62 4 35.1

8.63 6.13 14.2 8.82 13.2 6.64

2.23 1.75 4.18 2.84 3 74 2.28

215 204 316 231 250 210

6.02 5.58 8.21 6.53 6.78 5.22

0.278 0.375 0.202 0.304 0 255 0.318

90.6 59.7 63.8 75.9

14.0 10 7 13.6 13.7

3.92 3.22

365 279 300 339

9.88 7.64 7.22 8.62

0.234 0.293 0.267 0.270

319

795

0.314

Nd

3.56

711 20.8 24.1

S&m

0.219 0 275 0.427

44.2

9.65

72.7 38,5 42.2 94.3

11.7 9.40 8.34 17.5

3.64 2.B4 2.66 4.49

384 282 234 527

il.8 6.98 6.52 14.0

0.473 0.370 0 323 0 346

1.67 2.08

30.9 361

5.49 7.06

1.89 2.13

142 155

3.11 4.25

0.304 0.295

1.93

27.4

6.40

2 16

201

0 302

3588

K. P. Jochum. A. W. Hofmann, and H. M. Seufert Table 2~. Analytical

results of samples from the Hawaiian

Islands

(concentrahons

in

ppm) Sample KAUAI Kau-1 OAHlJ C 46 c 30 0A2 OA 7 0A5 OA 9 OA 11 OA 1 OA3 0A4 OA 5 0A6 OA 10 MOLOKAl G 44 71.WAIK 8F MAUI c 153 C 149 C 116 HMT 79-2b ct27 KAHOCIAWE KW-24 KW-25 KW-1 KW-2 KW-7 KW-6 KW-19 KW-18 H 1440 KW-14 HA WA// c 70 79MKl 1501 DAS-136 KL-2 ML-3B LO/H/ Dredge 2

Flock type

Sn

Nd

Sm

Eu

tholeiite

1.24

10.6

3.09

1.17

tholeiite alk. basalt tholeiite tholeiite tholeiite tholeiite tholeiite mell. neph. nephelinite nephelinite nephelinite alk. basalt alk. basalt

1.29 1.59 1.47 1.61 1.53 0.999 1.25 1 .96 1.70 1.60 1.50 0.941 1.21

20.6 30.7 20.7 23.0 24.7 16.0 21.8 68.6 54.7 48.6 78.9 22.2 26.7

5.74 7.04 5.15 6.64 5.70 4.21 5.31 13.1 11.0 9.19 17.0 4.70 5.76

alk. basalt thoieiite

2.77 1.59

49.6 18.3

10.7 4.52

3.40

hawarite mugearite trachyte alk. basalt hawaiite

2.14 4.60 5.12 2.60 2.46

39 9 72.1 75.5 26.5 43.9

tholeirte tholeirte tholeiite tholeiite tholeiite tholeiite tholeiite tholerite hawaitte hawaiite

0.590

1.17 1.19 1.23 1.53 1.46 1.36 I .20 2.84 2.05

alk. basalt ankaramrte alk. basalt tholeiite tholeiite tholeiite tholeiite

0.401

150 292 145 134 131 104 158 246 156 169 166 105 128

4.35 6.30 3.20 3.53 3.42 2.76 3.74 4.89 3.48 3.45 3.91 2.70 2.92

0.225 0.268 0.285 0.242 0.268 0.237 0.235 0.150 0.155 0 174 0.068 0.200 0 210

1.48

352 131

8.52 3.35

0.259 0.352

9.55 14.5 12.4 6.91 10.4

2.92 4.89 3.39 2.26 3.45

266 722 828 192 362

6.56 17.9 22.6 4.22 7.56

0.224 0.317 0.413 0.376 0.237

7.0% 10.5 15.2 16 3 20.4 13.5 21.4 17 5 61.4 45.7

2.15 2.55 4.18 4.45 5.94 4.10 5.25 4 02 15.2 129

0.802 1.04 1.57 1.45 2.21 1.99 1.38 4.32 4.31

112 111 164 175 234 142 178 131 462 313

2 60 2.72 3.86 4.16 5.72 3 75 3.75 3.20 10.2 7.14

0.414 0.459 0.285 0.276 0.308 0.356 0.259 0.299 0.187 0.159

2.93 1.61 1.34 I .53 1.89 1.63

47.5 21 2 24.2 18.6 22.3 17.2

10.2 4.96 4.67 4.53 5.78 4.56

3.07 1 65 I .64 1.68 1.98 1.69

409 155 148 121 163 134

9.22 2.98 3.14 3.54 4.35 3.42

0 287 0.325 0.287 0.317 0.327 0.335

1.19

17.2

3.74

1.32

136

3.40

0.318

RESULTS The tin analyses of terrestrial rocks are listed in Table 2ag. Because tin behaves like a moderateiy incompatible element (such as one of the middle REEs) during magmatic processes,

Table 2d. Analytical results (REE from WHITE and DUPRE, 1956) of IAV from the Lesser Antilles (concentrations in ppm)

KEJ 100 GDA 004 GDA 01 i WIG 020 MONT 102 SAB 102 GUAD 504 STL 207 1R STV 79-90

St-l Nd Sm ______ ~I______-________ 1.14 10.5 2.70 0.813 17.4 3.06 0.853 12.5 3.10 0.793 8.01 2.29 0.833 13.1 3.22 0.497 8.76 2.31 0.972 9.80 2.65 0.791 9.62 3.13 0.373 4.66 1.66 0.904 IO 1 3.04

Eu 0.946 0.930 I.03 0.543 1.14 0.834 0.521 1.17 0.610 1.05

SnlSm

2.15

measured tin concentrations in four carbonaceous chondrites of different types. These include the most primitive CI carbonaceous chondrites Orgueil and Ivuna.

Sample

Hf

Zr

Sn/Sm 0.422 0.266 0.275 0.346 0.259 0.215 0.367 0.253 0.225 0.297

1 93 2.32 1.70 1.97 1.52

1.41 1.73 4.63 3,97 3.09 5.65 1.81 1.9%

1.37

95.8

Table 2e. Analytical results (REE from JOCHUM et al., 1991) of Precambrian and Recent komatiites (concentrations in ppm) Sample

Sn

Nd

3.4 Ga ONl’ERWACHTGRoUp BARBERTON 5241 0.318 3.24 5243 0.212 1.33 5031 0.481 3.46 5019 0.252 2.30 2.7 Ga KAMBALDA A 1140 478 477 BELlNGWE 22 ZA I ZV 85 ALEX0 M 664 0.1

Sm

Eu

SniSm

1 04 0.410 0.832 0.621

0.323 0.144 0.305 0.233

0.306 0.517 0.578 0.406

0.196 0.302 0.243

1.49 2.06 2.13

0.567 0.505 0.761

0.236 0.252 0.255

0.346 0.375 0.319

0.276 0.274 0 222

2.14 1 98 2.01

0.792 0,772 0.658

0 291 0.320 0.244

0.348 0.355 0 323

0.229

1 .49

0.605

0.233

0.377

0.489 0.410

2.65 2.86

1.25 1.25

0.500 0.482

0.391 0.328

Ga

GOR 159 GOA 160

3589

Tin in mantle-derived rocks Table 21. Analyhcal results (REE from JOCHUM et al., 1991) of Precambrian and Recent basalts (concentrations in ppm) Sn

Sample

Nd

Sm

Table 2h. Sn concentrations chondrites of different types

(ppm)

in carbonaceous

Eu

SnfSm

Meteorite

Type

1.43 0.860 0.657

0.300 0.297 0.323

Orgueil-T

Cl

1.57, 1.64

1.61

Orgueil-318/l

Cl

1.62,

1.62

Orgueil-3i8/3a

Cl

1.70, 1.62

1.66

lvuna

Ci

1.60

1.60

Murchison

c2tl

0.97

0.97

Allende

c3v

0.53, 0.85, 0.59

0.59

mean

Sn

3.4 Ga ONVERWACHJGROUP, BARBERJON 24.3 5.30 5038 1.59 2.10 7.02 5080 0.624 2.04 7.25 5077 0.659 PlLBARA 5.50 1.75 92 0.577 4.44 13.9 56A 1.75 2.98 9.00 26A 1.16 2.7 Ga KAMBALDA KA 1 C 592

0.666 1.66 1.08

0.330 0.394 0.396

0.672 0.756

6.00 12.3

1.99 2.65

0.707 0.887

0.338 0.285

0.557 0.349

5.86 3 95

1.75 1.36

0.668 0.522

0.318 0.257

1.62 0.569 1 21 0 604

13.9 3.57 7 46 4.88

3.89 1.31 2.39 1.61

1.32 0.504 0.806 0.633

0.416 0.450 0.506 0.375

0.520 0.490

4.79 4.93

1.63 1.55

0.600 0.570

0.319 0.316

1.9 Ga OTTAWA ISLANDS G6 0.478 G 378 0.419 G 12 0.390

6.13 4.31 4.82

1.69 1.35

0.612 0.603 0.476

0.283 0.310 0.258

0.603 0.627 0.950

0.247 0.216 0.206

NEW 91 NEW 20 w Cl C 126 G 31 C6 WA WA WA 59 WA 75

0.1

I.51

(2.06’)

* Outlier is excluded from the mean

mantle heterogeneity and different degrees of partial melting and fractional crystallization. The tin content varies between the different rock types. For example, Hawaiian tholeiites average 1.4 + 0.3 ppm, alkali basalts 2.0 rt 0.8 ppm, and differentiated mugearites and trachytes up to 5 ppm (Table 2~). Samples from most oceanic islands have similar tin contents of about 2-3 ppm (Fig. 3). However, they are slightly higher in samples from Tristan da Cunha, Tubuai, and the Society Islands (up to about 5 ppm). These islands also have high concentrations of other incompatible lithophile elements.

Ga

G

167 0.355 GOR 117 0.376 GOR 54 0.530 _______---_------

2.90 3.53 9.51 ----_

1.44 1.74 2 57

Island Arc Voltanics island arc volcanics (IAV) from the Lesser Antilles have tin concentrations in the range of 0.4 to 1.1 ppm and a mean value of 0.8 ppm.

we have also included Nd, Sm, Eu, Zr, Hf abundances in this table. These were determined mainly by SSMS together with tin. Mid-Ocean Ridge Basalts Tin concentrations in MORB vary from 0.6 to 2.6 ppm (Table 2a) and correlate generally with the concentrations of the REE. Zr, and Hf. The average of all the tin analyses made on MORB glasses is Sn = 1.3 ppm.

Komatiites and Basalts (With Ages Ranging Up to 3.4 Ga) The mean tin concentration of all analyzed komatiites is 0.30 + 0.06 ppm (Table 2e). Tin abundances apparently do not vary systematically with the ages of these rocks. Tholeiitic basalts associated with komatiites have higher concentrations up to I .8 ppm (Table 2f). Peridotites

All the peridotite xenoliths measured were of the spinelbearing variety included in alkali basalts. There appears to

Oceanic Island Basalts The abundances of tin range from 0.9 to 6.1 ppm (Table 2b,c). This variation is caused by some combination of source AtI

Table 29. Analyttcal results (* from JOCHUM et al., 1969) of spine1 pertdotite xenoliths (concentrations in ppm) Sample

Location

Sn

Nd

0.140 0.140 0.210 0.093 0.110 0.052 0.363 0.092 0.104

1.35 0.619 0.911 0.409 0.468 0.448 5.39 1.80 0.887

------__.

SC-T’ PO-l’ Fr-1’ D-l’ KH-1’ 84.402 84-413 2669 85-l 68

San Carios Potrillo Landoz Dreiser Weiher Kilbourne Hole SE Australia SE Australia SE Australia SE Australia

Sm

Eu

---______

____________-_______--------_-----

0.441 0.298 0.341 0.142 0.233 0.123 1.13 0.271 0.237

0.170 0.113 0 119 0.050 0.087 0.039 0.350 0.070 0.073

LaA

Ga

P&HoR Ind

so

AGo O. Tr 0

Qn 00 C SH se

o

0

T”

SnlSm

--0.317 0.470 0.616 0.655 0.472 0.423 0.321 0 339 0.439

0.11 ” 0.6 0.8 I

2 Sn(ppm)

4

FIG. 3. Mean tin concentrations and Sn/Sm ratios of basatts from three major ocean basins (MORB), several ocean islands or island groups (OIB). and one island arc (IAV).

3590

K. P. Jochum, A. W. Hofmann. and Il. M. Seufert

be no systematic difference in tin abundance between the various locations. Overall, the mean tin abundance of all peridotite samples is 0.14 ppm. Continental Crust TAYLOR and MCLENNAN ( 1983) estimated a tin abundance of 2.5 ppm for the bulk continental crust. This value is derived from their “andesite model” using a large number of tin analyses of andesites. This included earlier literature data from ONISHI and SANDEL.L ( 1957; Sn = 1 ppm) and HAMACIUC’HIet al. ( 1964; Sn = 1.75 ppm), and an average value (Sn = 2.40 ppm) for I 19 andesites analyzed in the ANIJ laboratory. However, they calibrated their data using a theoretical calibration factor and recalibrated the literature data using their tin value for the G-l standard rock (5.1 ppm). As shown in Fig. 1. the tin data for standard rocks of TAYLOR and MCLENNAN ( 1983) are not well correlated with the reference values summarized by GOVINDARAJU ( 1989); this is in contrast to our ID-SSMS data (mean deviation of IDSSMS data from the reference values = 1.O + 0.2; see Table 1). Most of their values (including G- 1 ) are systematically higher of a factor of about 1.7 compared with the reference and our values. Therefore, we believe that the theoretically based recalibration in TAYLOR and MCLENNAN (1983) is not justified and. consequently, the continental-crustal abundance of tin may be 70% lower. Our best estimate is 1.5 ppm.

patibility empirically, we test global correlations with other purely Iithophile incompatible elements and search for the element having the most nearly constant ratio with tin. Formally, an element x that has the same bulk partition coefficient as tin in igneous processes (such as partial melting and fractional c~stallization) will have a correlation coefficient and slope equal to unity on a Sn vs. x concentration plot. On a Sri/// vs. Sn plot, on the other hand, the correlation coefficient and the slope will be zero. This procedure is equivalent to that used by NEWSOM et al. ( 1986) and HOFMANN et al. ( 1986). The latter authors showed that the slope criterion is a sufficient condition for demonstrating that the two elements have equal bulk partition coefficients, P and Do, during nonmodal partial melting, except for combinations of P and Do that are completely unrealistic in incompatible elements. NEWSOM et al. ( 1986) found that the moderately siderophile elements Pb, MO, and W have similar compatibilities as Ce, Pr, and Ba, respectively. In Fig. 4, the results for tin and the three neighbouring REEs, Nd. Sm, and Eu, in MORB and OIB samples are plotted. Nephelinites from the Honolulu volcanic Series (Oahu, Hawaii ) yield lower ratios of tin to these REEs. They have been omitted from Fig. 4 because these particufar samples also exhibit anomalies in other trace-element ratios (Zr/ Hf, Y/Ho, Zr/Sm; see JOCHUM et al., 1987). This may possibly

I 100

Sn/Nd

Carbonaceous Chondrites CI carbonaceous chondrites represent essentially unfractionated samples of solar system material for most refractory and volatile elements (e.g., PALME et al.. 1981; ANDERS and GREVESSE, 1989), and they are arguably the most suitable for determining solar system abundances. Three separate samples of the CI chondrite Orgueil were analyzed, yielding mean tin concentrations within the narrow range 1.61 to 1.66 ppm (Table 2h f. The mean Orgueil value of 1.63 ppm is 7% lower than an earlier SSMS measurement (tin = 1.75 ppm) reported by KNAB ( 198 1). High-precision measurements of tin using isotope dilution by thermal ionisation mass spectrometry (LOSSet al., 1989) give similar tin abundances in Orgueil, ranging from 1.57 to 2.04 ppm, with a mean value of 1.84 ppm. Our tin concentration ofthe CI chondrite Ivuna, I .60 ppm, is very similar to the Orgueil value. From these data. we obtain a mean value of I .62 + 0.03 ppm (20,) for Cf chondrites, which is 6% lower than the recommended value inferred by ANDERS and GREVESSE ( 1989). Analyses of the C20 and C3V chondrites Murchison and Allende give lower tin abundances in agreement with data for other moderately volatile elements. Our tin concentration for Allende is identical to the recently published value of059 ppm (LOSS et al., 1989) but 25% lower than the value reported by KNAB (1981). I~CO~~PA~IBILIT~

OF TIN

When no metal or sulfide phases are present, tin behaves like an ordinary (lithophile) incompatible element during igneous processes. To determine its relative level of incom-

3.1

----c--)--I-Iw Sn/Sm 1

D.1

1

10

I

Sn(ppmf

3.1

I

Sntppm)

FIG. 4. Nd, Sm, ELIvs. tin concentrations and ratios of tin with the three ne~ghbou~ng REEs for oceanic basalts (MORB: open squares. OIB: open circles.) Tin is best correlated with Sm (highest correlation coefficient r2, slope about 1. constant Sn/Sm ratios), indicating that tin and Sm must have similar bulk partition coefficients.

3591

Tin in mantle-derived rocks be due to the much lower degree of partial melting (
Zr/Hf

100

4 IO



La-

10

I

100

loo0

Zr(ppm)

Hf f ppm 1 /

snizr

‘-7

0,I

IO

/

1 01

1.6 ,

lW

0.6

’ 2

/

/

/

3

4

5

6

ionic charge FIG. 6. Plot of ionic radius vs. ionic charge. Moderately siderophile/ lithophile element groups which have constant ratios form subparallel arrays. The arrow shows the trend of the degree of incompatibility.

of incompatibility: that is, they have about the same bulk solid-liquid partition coefficients. Figure 5 shows that tin is also approximately equally as incompatible as Zr and Hf. Thus. Sn/Zr (0.01 1) and SniHf(0.4 I ) ratios are nearly uniform in both MORB and OIB. The geochemical behaviour of tin is consistent with the relationship between compatibility, ionic charge, and ionic radius (TAYLOR, 1964). This is demonstrated in Fig. 6. where the ionic radii of several moderately siderophile and lithophile elements are plotted vs. the respective ionic charges (WHITTAKER and MU~VTUS, 1970), together with contours ofequal compatibility in mantle-related igneous processes. Each group of elements sharing a common charge exhibits progressively decreasing bulk partition coefficients within the group as a function of increasing radius. The slopes of the contours show that bulk partitioning depends not only on ionic radius but also quite strongly on ionic charge. For a given radius. the partition coefficient decreases with increasing charge. In general, element groups of equal compatibility form subparailel arrays on this type of diagram (JOC~KJM et al., 1984). These relationships have been demonstrated previously for the moderately siderophile-lithophile element pairs W-Ba, PbCe, and Mo-Pr by NEWSOM et al. ( 1986). Tetravalent tin has an ionic radius of 0.77 A ( WHITTAKER and MUNTUS, 1970), very close to the moderately incompatible elements Zr (0.80 A) and Hf (0.79 A). The moderately incompatible behaviour of tin (comparable to Sm) is also consistent with the charge-radius trend seen in Fig. 6. If the previously established contours shown in Fig. 6 were used to predict the best REE compatibility analogue for tin (by drawing parallel contours), the “best” REE analogue would be the moderately incompatible element Ho. This element has an ionic radius of 0.98 A, which is close to that of the actual best analogue, Sm (1.04 A), as determined by the relationships shown in Fig. 4. CONSTRAINTS

FIG. 5. Relationships between Zr, Hf, and tin concentrations in oceanic basalts (MORB: open squares. OIB: open circles.) The diagrams show that tin is approximately equally as incompatible as Zr and Hf.

/

ON CORE ~ORM.~TlON

The tin data are consistent with and provide further confirmation for the constraints previously placed ( NEWSOM et al., 1986) on core formation and the possibility of late “core

3592

K. P. Jochum, A. W. Hofmann, and H. M. Seuferi

pumping” (ALL~GRE et al., 1982) as a possible explanation for the high U/Pb and Th/Pb ratios of the source materials of some OIB. This is particularly true of the “HIMU” type OIB of ZlNDLER and HART ( 1986). Examples of HIMU OIB included in this paper are St. Helena and Tubuai, which have (geometrical ) mean ratios of Sn/ Sm = 0.29 * 0.05 and 0.30 ? 0.09, respectively, neither of which differ significantly from the global means of Sn/Sm = 0.36 (MORB) and 0.3 1 (OIB). This confirms that these HIMU sources have not suffered large losses of moderately siderophile/chalcophile elements within the l-2 Ga time period that these sources have been in existence. We wish to clarify at this point that one conclusion of NEWSOM et al. ( 1986), namely that the analogous Mo/Pr and Pb/Ce ratios preclude late “core pumping” as a possible cause for the general p-increase of the mantle, is not fully justified by either the previous or the new data on recent basalts. If siderophile elements were extracted from the mantle later than initial core formation but prior to the differentiation of the various silicate reservoirs, this loss might be recorded simply as a secular change in the average Mo/Pr, Pb/Ce, or Sn/Sm ratios but not as a difference between different types of oceanic basal& Thus, MORB and OIB might have uniform values of these ratios, even if the core continued to grow during Archean time. What the data for modern basalts do show is that the source regions of ocean island basalts, which may be located near the core-mantle boundary, have not lost detectable amounts of siderophile elements to the core. Furthermore, we conclude that if the sources of plumes are located in the lower boundary layer of the mantle, then this boundary layer shows no evidence for chemical interaction with the core, recent speculations to the contrary ( KNITTLE and JEANLOZ, 1989) notwithstanding. A more general test of continued core growth during Precambrian time is provided by testing for secular changes in the Sn/Sm ratio. Figure 7 shows this for komatiites and basalts with ages up to 3.4 Ga. The Sn/Sm ratios do not vary with age in any systematic fashion. The ratios of the Precambrian rocks, with arithmetic means of 0.36, agree within the scatter of the data with the modern basalt values (MORB, Sn/Sm = 0.36; OIB, Sn/Sm = 0.31). Thus, there is no resolvable evidence for core growth during the last 3.4 Ga. Quantitative estimates of the maximum permissible amount of late core growth (either from OIB sources or from the mantle in general) must be taken with a grain of salt because neither the details of the hypothetical extraction mechanism nor the relevant metal/silicate partition coefficients are known sufficiently well. If we assume that a liquid metal phase segregates from the solid silicates in the lower mantle, we can estimate the degree of siderophile element depletion using the equilibrium equation of SHAW ( 1970), as follows: &/Co

= l/F(DML’=

-

1) + I),

where C, and C, are the concentrations of the residual and the initial solid, respectively; DMLjSSis the bulk partition coefficient between liquid metal and solid silicate; and F is the mass fraction of liquid metal. DMLfSS may be estimated to first order from the experimental partitioning data at 10 kb for tin between metal and silicate liquids, DMLISL = 27 ( SEI-

0.11

I 0

I

I 2

I 1

I

I 3

I 4

Age(Ga)

FIG. 7. Sn/Sm ratios for rocks of different ages.

FERT and RINGWOOD, 1987), and the partitioning of Sm between lower mantle silicate solid and silicate liquid, Dss’sL( Sm) = 0.2 (the result for Mg-perovskite of KATO et al., 1988). Assuming further that Dss’sL (Sn) = Dss/sL( Sm), as inferred from present data, and DMLtSS(Sn) = DMLiSL/ DsslsL, we finally obtain D MLiSS= 135. (The actual bulk partition coefficient for the lower mantle is expected to be similar, though certainly not identical, to that of Mg perovskite. It seems likely that the bulk value should be reduced by the presence of magnesio-wiistite but increased by the presence of Ca and Al in the perovskite). Inspection of Fig. 7 shows that a secuIar reduction of tin of more than 50% should be detectable by a corresponding decrease of the Sn/Sm ratio. For DMLISS(Sn) = 135, such a reduction of tin would be achieved by the equilibrium extraction of 0.75% of molten metal. If DML’SS(Sn) is arbitrarily reduced to a value of 50, the required amount of molten metal would be about 2%. We conclude that after core formation early in Earth history, the silicate portion ofthe Earth was extensively mixed by convection. This resulted in a homogeneous tin distribution in the silicate reservoirs of the Earth. This homogenization was completed prior to 3.4 Ga ago (the ok’ ;t rocks for which we have Sn-Sm data). The content of ‘n in the silicate portion of the Earth has not changed significantly since then. This indicates that the core did not grow during the last 3.4 Ga within the resolution afforded by the available tin, MO, Pb, and W data. PRIMITIVE

MANTLE ABUNDANCE

Estimates of the primitive mant. : abundance of tin vary widely from 0.1 to 0.7 1 ppm (e.g., NONAKA, 1982; ANDERSON, 1983; GANAPATHY and A~I\~~)ERs,1974; CHOU, 1978; SUN and MCDONOUGH, 1989). This is due mainly to the fact that mantle-derived rocks have only rarely been analyzed. In this study, the tin abundance of the primitive mantle is estimated using the two following independent methods: 1) We assume that the primitive mantle has been differentiated into different mantle reservoirs and continental crust. Figure 8 demonstrates that the mantle-derived igneous rocks have similar Sn/Sm ratios. We obtain a geometric mean Sn/Sm ratio of 0.316 -+ 0.011 (2a,) using log Sn /Sm values of 140 MORB, OIB, 1AV samples, as well as Precambrian and recent komatiites and basalts, which are normally distributed (Fig. 9). We argue that

Tin in mantle-derived

sn(PPm)

10

5

----I

CONT.CRUST

I

0.11

/‘,,I

1

0.1 gn ( ppm

10

1

FIG. 8. Sm and Sn/Sm ratios vs. tin concentrations derived rocks. The mean ratio agrees with the estimated

for mantleratio for the

continental crust. Tin abundances and Sn/Sm ratios for the primitive mantle (PRIMA) and for the CI carbonaceous chondrites Orgueil and Ivuna are also shown.

this mean

ratio is representative of the Earth’s is because estimates based upon the continental-crustal abundance of Sn ( 1.5 ppm) and Sm (4.4 ppm; see TAYLOR and MCLENNAN, 1985; WEDEPOHL, 198 1; WEAVER and TARNEY, 1984) yield a similar ratio (0.34). Thus, in contrast with the situation found for Nb/U and Ce/Pb, which are very different in the continental crust than in mantle-derived rocks (HOFprimitive

3593

MANN et al., 1986), the Sn/Sm ratio appears to be identical in all available silicate reservoirs. Consequently, taking Sm = 0.387 ppm for the bulk silicate Earth (HOFMANN, 1988), we obtain Sn = 0.12 + 0.01 ppm for the primitive mantle. 2) A primitive mantle composition can also be estimated from the chemistry of fertile spine1 peridotite xenoliths. In this respect, an important sample is the nodule SC- 1, which is “primitive” with respect to its major element and compatible trace-element chemistry ( JAGOUTZ et al., 1979). SC- 1 has a tin concentration of 0.14 ppm. In order to calculate a primitive mantle abundance from this, however, the crustal contribution to the tin budget must be added back in, as described by W~~NKE et al. ( 1984). This contribution is small for the moderately incompatible element Sn (0.009 ppm), assuming that the continental crust comprises 0.60% of the total silicate portion of the Earth (DAVIES, 198 1). Taking this into account, we obtain a primitive mantle abundance of 0.15 ppm tin.

0 MORB o OIB d IAV l Komatiites and Basaltr A Peridotite Xenoliths

E ?

rocks

The primitive mantle abundance of tin can also be estimated from the Sn-Sm relationship of all peridotite xenoliths (Fig. 10). Tin is well correlated with Sm. Assuming a primitive mantle abundance of Sm of 0.387 ppm (HOFMANN, 1988 ), we obtain 0.15 f 0.02 ppm tin for the primitive mantle. It is reassuring that both estimates derived from the chemistry of spine1 peridotite xenoliths are identical within error limits to those obtained from the constant Sn/Sm ratios in mantle-derived volcanic rocks.

Sn/Sm

mantle.

This

DEPLETION

IN THE PRIMITIVE

MANTLE

All the moderately siderophile elements are depleted in the primitive mantle relative to solar system abundances. The tin depletion factor is 33 f 3 (Fig. 8). This is derived from the Sn/Sm ratios of 0.3 16 and 10.5 for mantle-derived rocks and CI chondrites, respectively (Table 2h; Sm = 0.154 ppm for Orgueil and Ivuna, also analyzed by ID-SSMS together with tin). This value is very similar to the corresponding depletion factors for W (26 f 9) and MO (42 + 5 ) obtained in a similar manner by NEWSOM et al. (1986). However, recent data by SIMS et al. ( 1990) yield a considerably lower W depletion value (about 10) for the primitive mantle, assuming that the depleted and the continental crustal reservoirs have significantly different ratios.

3

log(Sn/Sm)

SnlSm

FIG. 9. Distribution of log,,(Sn/Sm) and Sn/Sm ratios for mantlederived igneous rocks (shaded regions) and spine1 peridotite xenoliths (open regions). As can be seen, the Sn/Sm ratios are not normally distributed, as opposed to the log,,(Sn/Sm) distribution. We have therefore used the latter geometric mean rather than the arithmetic mean for estimating the mantle Sn/Sm.

0.5

1.0

Sm(ppm)

FIG. IO. Tin vs. Sm abundances for spine1 peridotite xenoliths. The primitive mantle (PRIMA) abundance of tin can be estimated from this relationship.

K. P. Jochum, A. W. Hofmann. and H. M. Seufert

3594

An important point to note is that the moderately siderophile elements Sn, W, and MO differ not only in their partition coefficients but also greatly in their volatility. For example, W is a refractory element having a 50% condensation temperature of 1802K ( WASSON, 1985), whereas tin (condensation temperature = 720K) is one of the most volatile elements of the entire group of moderately siderophile elements. NEWSOM ( 1990) investigated moderately siderophile element depletions in the Earth’s mantle as a function of volatility. He showed that after volatile correction (by subtracting out the depletion of volatile lithophile elements of similar volatility), most of the moderately siderophile elements are depleted by factors of 5-20. These values represent the depletion only to their siderophile behaviour. Using the correction of NEWSOM ( 1990)) we obtain a volatilty corrected tin depletion of about 2-10, which overlaps the results obtained for other moderately siderophile elements. The large error of the tin depletion factor is caused by the lithophile element trend, which is not well defined at the low condensation temperature of tin. This observation must be satisfied by any comprehensive model of accretion and core formation in the early Earth. As has previously been shown by WANKE et al. (1984), moderately siderophile elements have very similar depletion factors in spite of their grossly contrasting metal/silicate partition coefficients. Our result for tin is in general agreement with the inhomogeneous, two-component accretion model of W.XNKE ( 198 1). in which the second accretionary stage added the moderately siderophile elements in chondritic abundances after segregation of metal to the core had taken place. The arguments regarding accretion advanced here, which essentially follow and reinforce those of WANKE ( 198 1) should nevertheless be viewed with some caution because the pressure dependences of the relevant partition coefficients, as well as actual pressures relevant to core segregation, are far from well known. CONCLUSIONS

1) Sn/Sm

ratios are approximately uniform (within a factor of 2) in different mantle reservoirs (MORB, OIB sources). They agree, within the data scatter, with those found in Precambrian rocks. These results suggest that the core did not grow during the last 3.4 Ga. 2) Our results further indicate that tin in the mantle has been homogenized after core formation. This homogenization was completed 3.4 Ga ago. 3) The present-day mantle Sn/Sm ratio of0.32 appears also to be the primitive mantle ratio because estimates of the continental-crustal abundances yield similar values. We obtain Sn = 0.12 ppm for the primitive mantle, which agrees well with estimates determined from the chemistry of fertile spine1 peridotite xenoliths (Sn = 0.15 ppm). 4) Tin is depleted in the primitive mantle relative to CI chondritic abundances by a factor of 33. This depletion is due to both volatility and siderophility (NEWSOM, 1990). The volatility corrected depletion factor of tin is similar to other moderatey siderophile elements in agreement with heterogeneous accretion ( WXNKE, 198 1). ..1c.Xnorl,/t,u’qr?lcnt.v-Thesamples used in this study have been accumulated over the course of many years. They have been contributed

by many individuals. several of whom have been coauthors in several previous publications. particularly E. Ito, W. M. White. N. T. Amdt, L. M. Echeverria, W. F. McDonough. and H. Palme. In addition, some of the Hawaiian samples have been collected by AWH, but the majority have been contributed by M. Tatsumoto and by W. P. Leeman. S. Midinet-Best assisted during SSMS analyses, and S. Galer provided useful comments on the manuscript. We appreciate constructive reviews by H. E. Newsom. J. Ruiz. and C. R. Neal.

Editorial hurdling: H. E. Newsom REFERENCES ALL~GREC. J., DUPREB.. and BREVART0. ( 1982) Chemical aspects of the formation of the core. Phi/. Trans. Roy. Sot. London A306, 49-59. ANDERSE. and GREVESSEN. ( 1989) Abundances of the elements: Meteoritic and solar. Geochim. C‘osmochim. Ada 53, 191-214. ANDERSOND. L. ( 1983) Chemical composition of the mantle. Proc. 14th LWKU Planet. Sci. Cor$, Purt I; .J. Geophys. Rex 88, B4lB52 (suppl.) CHOU C.-L. (1978) Fractionation of siderophile elements in the Earth’s upper mantle. Proc. 9ih Lunur P/awl. Sci. Corzf. pp. 2 19230. DAVIESG. F. ( 198 1) Earth’s neodymium budget and structure and evolution of the mantle. Nalzfre 290, 208-2 13. GANAPATHYR. and ANDERSE. ( 1974) Bulk compositions of the moon and earth, estimated from meteorites. Prproc.5th Lunar Cwzf.‘, Vol. 2, pp. 1181-1206. GOVINDARAJUK. ( 1989) 1989 compilation of working values and sample description for 272 geostandards. Geo.ytund. Newlett. 13, l-l 13. HAMAGUCHIH., KURODA R., ONUMA N.. KAWABUCHIK., MITSUBAYASHIT.. and HOSOHARAK. (1964) The geochemistry of tin. Geochim. Cosmochrm. At/u 28, 1039-1053. HEDGEC. E., PETERMANZ. E., and DICKINSONW. R. (1972) Petrogenesis oflavas from Western Samoa. GSA Bdl. 83,2709-27 14. HEGNERE.. UNRLJHD.. and TATSUMOTOM. ( 1986) Nd-Sr-Pb isotope constraints on the sources of West Maui volcano, Hawaii. Nature 319, 478-480. HOFMANNA. W. ( 1988) Chemical differentiation of the Earth: The relationship between mantle, continental crust. and oceanic crust. Earth Planer. Sci. Let!. 90, 297-3 14. HOFMANNA. W., JOCHUMK. P., SEUFERTM.. and WHITE W. M. ( 1986) Nb and Pb in oceanic basalts: New constraints on mantle evolution. Earlh Planet. Sci. Let/. 79, 33-45. ITO E.. WHITEW. M., and GOPELC. ( 1987) The 0, Sr, Nd. and Pb isotope geochemistry of MORB. Chem. Geol. 62, 157- 176. JAGOUTZE., PALMF.H., BADDENHA~JSEN H.. BLUMK., CENDALES M.. DREIBUSG.. SPETTELB., LORENZV., and WWNKEH. ( 1979) The abundances of major. minor. and trace elements in the Earth’s mantle as derived from primitive ultramafic nodules. Proc. 1Olh Lunur P/awl. Sci. Corzf, pp. 203 l-2050. JOC‘HUMK. P. and VERMAS. P. ( 1993) Antimony in oceanic basalts: Indicator for seawater alteration. Terra Ah.str. Suppl. No. 7 10 Tow NOW 5, 342 (abstr.). JOCHUMK. P., HOFMANNA. W., ITOE.. SEUFERTH. M., and WHITE W. M. ( 1983) K, U. and Th in mid-ocean ridge basalt glasses and heat production K/U and K/Rb in the mantle. Nulwc306,43 I436. JOcHuM K. P.. HOFMANNA. W.. and SE~JFERTH. M. ( 1984) Global trace-element systematics in oceanic basalts. 27th Id Gcol Con~rc’.\s,,2foscow. .Ahsfr. Vol. IX, Part 2, p. 190 (abstr.). JOCHUMK. P.. HOFMANNA. W.. and SEUFERTH. M. (1985) Sn and Sb in oceanic basaltsand the depletion ofsiderophile elements in the primitive mantle. Eos 66, I I I3 (abstr.) JOCHUMK. P.. SELJFERTM., HOFMANNA. W., UNRUH D. M.. and STILLEP. ( 1987) Non-primitive mantle sources for Hawaiian basalts. Eo.s 68, 1549 (abstr.).

JOctiuM K. P.. SEUFERTH. M.. MIDINET-BEST S.. RETTM~NNE.. SCH~NBERGER K.. and ZIMMERM. ( 1988) Multi-element analysis by isotope dilution-spark source mass spectrometry (ID-SSMS). Fwtcniuc L. lnui. CJwn. 331, 104-l 10. JOCHUMK. P.. McDONOuc;ti W. F.. PAI~MEH.. and SPETTELB.

3595

Tin in mantle-derived rocks ( 1989) Compositional constraints on the continental lithospheric mantle from trace elements in spine1 peridotite xenoliths. Nature 340, 548-550. JOCHUM K. P., ARNDT N. T., and HOFMANN A. W. ( 1991) Nb-ThLa in komatiites and basalts: Constraints on komatiite petrogenesis and mantle evolution. Eu& Pkunri. Sfi. L&t. 107,272-289. KAYOT., RINGWOODA. E., and IRIFUNET. ( 1988) Constraints on element partition coefficients between MgSiOl perovskite and liquid determined by direct measurements. E~hrtlrPlunrt. Sci. Let/. 90,65-68. KNABH.-J. ( 1981) The distribution oftrace elements in carbonaceous chondrites. Geocllitn. C’osn7c~~l1im. Acta 45, I563- I572 KNITTI.I~E. and JEANLOZR. ( 1989) Simulating the core-mantle boundary: An ex~~mental study ofhigh-pinup reaction between silicates and liquid iron. ~~~~~~~~~~~. .&s. Lett. 16. 609-6 12. Loss R. D.. ROS~~ANK. J. R.. and DE LAETERJ. R. (1989) The solar system abundance of tin. Gcwchim. Co.smochim. Acts 53, 933-935. M~DONOUGH W. F. and MCCULLOCHM. T. ( 1987) The southeast Australian lithospheric mantle: Isotopic and geochemical constraints on its growth and evolution. &rtlr Pluw. Sc/. Lett. 86, 327-340.

MELSON W. G., BYERLYG. R., NELENJ. A., O’HEARNT.. WRIGHT

T. L.. and VALL~ERT. ( 1977) A catalog of the major element chemistry of abyssal volcanic glasses. Srn;t~?,s~)n;fin Contrrh. Earth .%i. 19, 3 l-60.

NEWSOMH. E. ( 1990) Accretion and core formation in the Earth: Evidence from siderophile elements. In Origin ofthc Eurth (ed. H. E. NEWSOM and J. H. JONES). pp. 273-288. Oxford Univ. Press. NEWSOMH. E. and PALMEH. ( 1984) The depletion of siderophile elements in the Earth’s mantle: New evidence from molybdenum and tungsten. Eurih Planc>t Sci. Left. 69, 354-364. NEWSOM H. E., WHITE W. M., JOCHLJMK. P., and HO~A~N A. W. ( 1986) Siderophile and chalcophile element abundances in oceanic hasalts, Pb isotope evolution, and growth of the Earth’s core. &~th Plawt. Sci. Letl. 80, 299-3 13. NONAKAJ. ( 1982) Uber die Haufigkeit von bisher wenig untersuchten Elementen im Erdmantel. Ph.D. thesis, Universitat Mainz. 0~1.~~1H. and SANDELLE. B. (1957) Meteoritic and terrestrial abundance of tin. Gwchim. Cosmc&zim. Acta 12, 262-270. PALMEH., SUESSH. E., and ZEH H. D. ( 1981) Abundances of the elements in the solar system. In ~u~~u~f-~~j~~s~ej~,GuntcpIIf: AStronom1~ .,~str~p~~~.s~~~und Spaw R~,~i~ur~~. Voi. 2; ‘4str~~n~~~~~nnd Astrophwic~. E&Won and Sq$. to Vu/. 1. Szdwol. A (ed.

K.-H. HELLWEGE),pp. 257-272. Springer-Verlag. PUCHELT-H. and EMMERMANNR. ( 1983) Petrogenetic implications of tholeiitic basalt glasses from the East Pacific Rise and the Galapagos Spreading Center. Chem. Geol. 38, 39-56. SEIFERT S. and RINGWOODA. E. ( 1987) Metal-silicate partition coefficients for some volatile siderophile elements and implications for lunar origin. Lamar Pkzncx Sci. XVIII, pp. 904-905.

SHAW D. M. ( 1970) Trace-element fractionation during anatexis. Geochim.

Cosmochim.

Aeta 34, 237-243.

SIMSK. W. W.. NEWSOMH. E., and G~ADNEVE. S. ( 1990) Chemical fractionation during formation ofthe Earth‘s core and continental crust: Clues from As. Sb. W. and MO. In Or&in ~~~.i~l~~ Eurih (ed. H. E. NEWSOM and J. H. JONES). pp. 291-317. Oxford Univ. Press. STILE P.. UNRUH D. M.. and TA~SJMOTO M. (1983) Pb. Sr. Nd. and Hf isotonic evidence of multiple sources for 0aht.r. Hawaii basalts. &‘&c;e 304, 25-29. STILLEP., UNRLJII D. M.. and TATSUMOTOM. ( 1986) Pb. Sr. Nd. and Hf isotopic constraints on the origin of Hawaiian basalts and evidence for a unique mantle source. ~~‘~~~~~~F?7. ~(~,S}~?O~/~~~~?. .ktU 50,2303-23 19. S~JNS.-S. and M~L~WOUGH W. F. (1989) Chemical and isotopic systematics of oceanic basalts: Implications for mantle composition and processes. In ,llu~~wmti.tm iti the OUWI Bt/.si~ (ed. A. D. SA~JNDERS and M. J. NORRY ):licrrl SOC Land. Sprc,. Puhi..pp. 313-345. TAYLORS. R. ( 1964) Trace-element abundances and the chondritic Earth model. Geochim. Co.smochim. Actu 28, 1989-I 998. TAYLORS. R. and MCLENNANS. M. ( 1983) Geochemical application ofspark-source mass spectrography. IV: The crustal abundance of tin. C%c~r. G&l. 39, 273-280. TAYLORS. R. and MCLENNAN S. M. ( 1985) T!rc C~)ni~n~~~~tulCrust: ft.7 C’o~77posirion und Ew&ion. Blackwell. WANKE H. ( 198 1) Constitution of terrestrial planets. Phil. Trans. Roy. Sot. London

A303,

287-302.

H.. DREIBUSG.. and JA~O(~TZE. ( 1984) Mantle chemistry and accretion historv of the Earth. In Archueun Gcochemiwt~ . (ed. A. KRONER et al.), pp. l-24. Springer-Verlag. WASSONJ. 7. ( 1985 ) M&ori/ex Freeman. WEAVERB. and TARNEYJ. ( 1984) Empirical approach to estimating the composition of the continental crust. .%ztur~ 310, 575-577. WEDEPOHLK. H. (1981) Der primare Erdmantel und die durch Krustenbildung verarmte Mantelzusammensetzung. Fortschr. Minwd 59 (Beiheft), 203-205. WEST H. B., GERLACHD. C.. LEEMANW. P.. and GARCIAM. 0. ( 1987) Isotopic constraints on the origin of Hawaiian lavas from the Maui Volcanic Complex, Hawaii. Nature330, 216-220. WHITE W. M. and DUPRE B. (1986) Sediment subduction and magma genesis in the Lesser Antilles: Isotopic and trace-element constraints. J G‘euphv.y. Res. 91, S927-S94 1. WI-IITEW. M.. TAPIA M. D. M., and S~HILLINGJ.-G. ( 1979) The petrology and geochemistry of the Azores Islands. Cow&. M%wd. Petro/. 69, 20 l-2 13. WHITTAKERE. J. W. and MUNT~JSR. ( 1970) Ionic radii for use in geochemistry. &whim. Cosmochim. ilctu 34, 945-956. WRIGI-ITT. L., SWANSOND. A., and DUFFIELDW. A. ( 1975) Chemical compositions of Kilauea east-rift lava, 1968- 1971. J. Prtwl. 16, 110-133. ZINDLIEK A. and HART S. R. (1986) Chemical geodynamics. Ann. Rev. fkrth Plwwt. Sci. 86 ( 14 ) ~493-570. WANKE